Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/plasma-ai/fractal/providersgit clone --depth 1 https://github.com/plasma-ai/fractalWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/plasma-ai/fractal/providers)<a href="https://agentmods.dev/agents/plasma-ai/fractal/providers"><img src="https://agentmods.dev/badge/agents/plasma-ai/fractal/providers.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00004 | $0.00990 |
| Opus 5 | $0.00002 | $0.00495 |
| Sonnet 5 | $0.00001 | $0.00198 |
| Haiku 4.5 | $0.00000 | $0.00099 |
Grade A, and why
features/agents/providers scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 85 lines — stays where its author put it; the contents beside it link to each section on GitHub.
features/agents/providers
[[features/agents/_index|..]]
Supported backends
Five provider backends ship in fractal/impl/, one module per provider, each
registered in fractal/core/agent.py under its base command word: claude,
codex, grok, opencode, and omp. The base command is the backend's
identity everywhere: the executable word, the DB agent column, the session-map
key, and the stem of the node's per-agent config directory (e.g. .claude).
An agent owns its own conversation loop inside a provider subprocess. The base-class contract is capabilities — invocation building, spawning, stream parsing, session policy, cost mode — never a message list. Backends differ along declared capability attributes rather than code paths: whether the session can be forked; whether the agent mints its own session id (codex) or the caller mints and pre-stamps it (claude); whether cost figures are per-invocation or cumulative thread totals settled against prior sibling steps; whether the backend needs LiteLLM token pricing (fatal at run start if the fetch fails); whether it accepts a hard per-step budget flag; and whether it emits one result frame per step (opencode) rather than one terminal result per turn.
The registry
fractal/core/agent.py holds the registry mapping each base command to its
backend — either a module:Class string imported lazily at resolve time, or a
class object. Core reaches impl/ only through this registry, so backend
imports stay off every code path that never spawns an agent. The registry has
three public verbs:
- resolve returns the backend class for a base command, raising a
ValueErrornaming the supported commands when the name is unregistered or its module cannot be imported. - register injects or overrides a backend under a base command — the in-process extension point for an embedding application. Explicit registrations win over hook-file registrations.
- supported lists the registered base commands in registration order.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 85 lines · 4 tokens per session scan A 1c536ed9a70f
features/agents/providers is an agent published in the GitHub repository plasma-ai/fractal (704 stars, last pushed 4d ago), licensed Apache-2.0. It adds 4 tokens to every session and 990 once invoked, about $0.0000 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
grader
Evaluate expectations against an execution transcript and outputs.
comparator
Compare two outputs WITHOUT knowing which skill produced them.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.